Question 20/12 – Objective assessment of conversational speech quality in networks
(new question)

Motivation

The telecommunications industry is working to adopt more flexible infrastructure to control costs and facilitate the introduction of new services. Examples are next generation IP networks (NGN) and 3G mobile networks – both of which exhibit time-varying performance. Measures that predict user-experience are useful in monitoring and managing time-varying performance and help to facilitate the rollout, efficient operation and effective service management of such networks. 

The industry is already benefiting from recent and proposed ITU-T recommendations for objective speech listening-quality assessment. However, typical communications involve interactive, two-way, conversations. IP and mobile networks can be particularly deleterious to interactive applications, including voice conversation; for example due to increased delay, which in turn will increase the probability of double-talk and increase the perceptibility of echo.

There is therefore an immediate need for a real-time, or near real-time, conversational speech quality assessment method. Such a method could combine ‘conversational’ impairments, such as level, echo and delay, together with ‘listening quality’ measures to provide an assessment of the overall (conversational) quality perceived by the user at either end of the connection.

It is also envisaged that such a method could use the outputs from existing objective listening quality models (e.g. P.862, P.563, together with information on ‘conversational’ network measures (such as those in P.561) to assess the overall conversational quality perceived by the users.
It is envisaged that such a method would be developed collaboratively. 

Question
Study items to be considered include, but are not limited to: 

• What existing models and measures could be used as inputs to the new method?
  
• What are the critical components of conversational speech quality
  
• What existing parametric and psychophysical models, or components thereof, could be used as building blocks for the new real-time method?
  
• What subjective test methods should method validation be based on?
  
• How can listening quality and conversational parameters be combined to assess overall perception of conversational quality?
  
• What subjective test data is needed to develop the new method?
  
• What additional considerations are relevant for wideband speech applications?

Tasks
Tasks include, but are not limited to:

• Define the scope of P.CQO, which will combine multiple objective measurements to provide an assessment of the perceived conversational speech quality in networks
• Review existing methods and models, and identify missing components
  
• Define inputs and outputs of P.CQO
  
• Identification of relevant subjective test methodologies (in co-operation with Q.7/12)
  
• Identification of existing subjective test data
  
• Collection of new subjective test data
  
• Development of new Recommendation P.CQO on assessment of conversational speech quality
  
• Ensure consistency with development of P.OQN (Q.6/12)
  

Relationships

Recommendations: P.340, P.561, P.562, P.563, P.800, P.800.1, P.831, P.832, P.834, P.862, G.107, G.108

Questions: 2, 6, 7, 8, 9, 10, 11, 13, 14, 17/12

Other study groups: SG2, SG 9, SG 16

Standardisation bodies: ETSI STQ, IETF